线性回归模型多个离群点的向前逐步诊断方法

当线性回归模型中存在多个离群点时,经典的诊断方法常常因掩盖和淹没现象而失效,导致模型误用。针对此问题,本文在回顾有关文献的基础上,将稳健回归技术与经典诊断量相结合,提出一种向前逐步诊断方法。通过对模拟数据的分析,说明该法可有效地识别回归数据中潜在的离群点,并作正式的统计检验。     

回归变量选择中的数据诊断

在多元线性回归中,变量选择紧密依赖模型,与影响数据密切相关。本文从模型扰动的角度,研究了变量选择与数据的关系,用微分几何中的概念,提出了用曲线的变化率、加速率及其曲率三种量测,去评价数据对变量选择的影响,从而诊断影响数据。文中给出的数值例子表明,所提影响量测,对于诊断数据对变量选择的影响是有效的。     

用时间序列方法预测股票价格初探

本文提出了描述股价变动的一种模型,讨论了模型建立、模型预测的一整套方法及改进算法,比较了算法与改进算法间的优劣,并通过实证分析说明整套理论有一定的可行性。大部分在本文中讨论的算法对于可用核模型描述的其他时间序列的预测问题也同样适用。     

一种半参数ROC曲面估计方法

ROC曲面是诊断医学统计学里评估有多类诊断结果的诊断测试方法准确性的一个重要工具,也是近年来的一个研究热点.本文提出一种半参数的ROC曲面估计方法.该方法可以借助于许多统计软件里的逻辑斯蒂回归程序进行计算,所以它的实施较为方便.相关统计模拟显示,本文提出的方法与传统的非参数方法相比,有效性得到了显著提高.而与参数方法相比,当参数模型假设是正确时仍比参数方法有略高的有效性；而当参数模型假设不正确时,本文提出的半参数方法明显优于参数方法.     

回归诊断中几种影响诊断量的比较

在回归诊断的实际应用中，人们越来越认识到强影响点对回归模型的影响重要性。但由于影响评价的方法越来越多。使应用者不知采用何种方法为好。本文对一些较为常用的诊断量作了比较，并通过实例考虑它们的应用情况。     

关于回归模型的参数估计效率

本文讨论回归模型的参数估计效率。本文说明了现有线性回归模型的参数估计效率的下界与真实的参数估计效率在很多情况下相差较大,而且这种下界对于实测数据处理很难得到精确值。本文给出了估算参数估计效率的仿真方法。理论分析表明,该方法给出的参数估计效率的估计较现有的下界估计更合理;仿真和实算结果表明,对于一大类线性和非线性回归模型,该方法给出的回归模型的参数估计效率的估计更接近模型参数估计效率的真值。     

PLS回归在国际机场服务改进中的应用

为了帮助国际机场在资源有限的情况下,明确服务改进方向从而实现服务质量的改善,本文在现有调查数据的基础上,建立了一个能够明确机场各项服务对“旅客总体满意度”直接影响程度的模型,并给出了运用此模型的诊断思路和方法。鉴于建模目的需要以及对消除多重共线性方法的对比分析,本文选用了偏最小二乘(PLS)回归分析方法。最后对其中一个国际机场进行了诊断,指出了机场服务改进的方向。     

基于Pena距离的偏正态数据下位置回归模型的统计诊断

Pena距离是研究偏态数据的一种有用工具.本文利用Pena距离研究了偏正态数据下位置回归模型的统计诊断问题,得到了位置回归模型下Pena距离的表达式,对其性质进行讨论,从而得到高杠杆异常点的判别方法. Pena距离与Cook距离、似然距离进行比较,得到在一定的条件下Pena距离优于Cook、似然距离.通过随机模拟试验研究和实例分析,表明本文提出的理论和方法是科学合理的.     

基于统计方法的神经网络预测模型研究

与现有网络结构设计方法不同,本文将RBF网络解释为解释变量和被解释变量之间的一个非线性函数,基于RBF网络的学习动态特性提出2种修剪模型WRBF和TRBF。这两种模型根据参数显著性增加和删减节点,为网络结构设计提供了理论依据。对中国信贷序列预测的结果表明,这些模型能够识别外移、萎缩和衰减等冗余核函数,得到的精简网络具有最好的预测精度,对于提高货币政策前瞻性具有很好应用价值。     

偏正态数据下众数回归模型的参数估计及应用

本文针对金融、经济、社会科学、环境科学、工程技术和生物医学等研究领域存在的不对称数据,提出偏正态数据下众数回归模型,基于牛顿-拉弗森迭代利用EM算法来估计未知参数。通过Monte Carlo模拟和BMI数据实例分析验证,表明本文所提出方法的有效性,对于偏正态数据众数回归模型的估计效果优于均值回归模型。     

Logistic回归模型的统计诊断

统计诊断的主要任务就是通过诊断统计量检测已知观测数据在用既定模型拟合时的合理性,主要是找出数据当中的异常点或强影响点。本文主要研究Logostic回归模型的诊断统计量和诊断统计图。用牛顿迭代法给出Logistic回归模型的极大似然估计值,根据扰动模型得到传统的诊断统计量,结合残差、杠杆值和系数变化三者构造新的诊断统计量,绘制新的诊断统计图,通过模拟研究说明新的诊断统计量的有效性,最后用一个实际案例说明新的诊断方法的应用并进一步验证其优越性。     

一类相合的分位点型检验

广义非参数似然比检验统计量是一类很广的统计量,包含了众多重要的检验统计量,如Anderson-Darling(AD)等.利用Rubin的随机经验分布函数替代经验分布函数的方法,得到了广义非参数似然比检验统计量的新版本,构造了新的检验统计量.由于新的检验统计量在给定样本下仍然是随机变量,选择了它的分位点和期望作为检验统计量,分别称之为分位点型检验统计量和期望型检验统计量.在简单假设情况下,证明了分位点型检验统计量和期望型检验统计量在固定备择下的相合性.模拟结果显示,在某些备择下,新的检验的功效明显高于原有的基于经验分布函数的检验的功效.     

On Exponential Representations of Log-Spacings of Extreme Order Statistics

In Beirlant et al. (1999) and Feuerverger and Hall (1999) an exponential regression model (ERM) was introduced on the basis of scaled log-spacings between subsequent extreme order statistics from a Pareto-type distribution. This lead to the construction of new bias-corrected estimators for the tail index. In this note, under quite general conditions, asymptotic justification for this regression model is given as well as for resulting tail index estimators. Also, we discuss diagnostic methods for adaptive selection of the threshold when using the Hill (1975) estimator which follow from the ERM approach. We show how the diagnostic presented in Guillou and Hall (2001) is linked to the ERM, while a new proposal is suggested. We also provide some small sample comparisons with other existing methods.     

Testing for Varying Dispersion in Exponential Family Nonlinear Models

A diagnostic model and several new diagnostic statistics are proposed for testing for varying dispersion in exponential family nonlinear models. A score statistic and an adjusted score statistic based on Cox and Reid (1987, J. Roy. Statist. Soc. Ser. B, 55, 467-471) are derived in normal, inverse Gaussian, and gamma nonlinear models. An adjusted likelihood ratio statistic is also given for normal and inverse Gaussian nonlinear models. The results of simulation studies are presented, which show that the adjusted tests keep their sizes better and are more powerful than the ordinary tests.     

General relations and identities for order statistics from non-independent non-identical variables

Some recurrence relations and identities for order statistics are extended to the most general case where the random variables are assumed to be non-independent non-identically distributed. In addition, some new identities are given. The results can be used to reduce the computations considerably and also to establish some interesting combinatorial identities.     

An improved model averaging scheme for logistic regression

Recently, penalized regression methods have attracted much attention in the statistical literature. In this article, we argue that such methods can be improved for the purposes of prediction by utilizing model averaging ideas. We propose a new algorithm that combines penalized regression with model averaging for improved prediction. We also discuss the issue of model selection versus model averaging and propose a diagnostic based on the notion of generalized degrees of freedom. The proposed methods are studied using both simulated and real data.     

A martingale approach to scan statistics

Scan statistics are commonly used in biology, medicine, engineering and other fields where interest is in the probability of observing clusters of events in a window at an unknown location. Due to the dependent nature of the number of events in a large number of overlapping window locations, even approximate solutions for the simplest scan statistics may require elaborate calculations. We propose a new martingale method which allows one to approximate the distribution for a wide variety of scan statistics, including some for which analytical results are computationally infeasible.     

Multiply-refined enumeration of alternating sign matrices

Four natural boundary statistics and two natural bulk statistics are considered for alternating sign matrices (ASMs). Specifically, these statistics are the positions of the 1’s in the first and last rows and columns of an ASM, and the numbers of generalized inversions and −1’s in an ASM. Previously-known and related results for the exact enumeration of ASMs with prescribed values of some of these statistics are discussed in detail. A quadratic relation which recursively determines the generating function associated with all six statistics is then obtained. This relation also leads to various new identities satisfied by generating functions associated with fewer than six of the statistics. The derivation of the relation involves combining the Desnanot–Jacobi determinant identity with the Izergin–Korepin formula for the partition function of the six-vertex model with domain-wall boundary conditions.     

图模型及其在统计建模中的应用

图模型是近年来兴起的一个新的统计研究方向,主要借助拓扑图的直观形式对多维概率分布进行统计推断和因果分析。本文对图模型的基本概念、原理、建模及应用进行了简单介绍。     

Influence curves of general statistics

The influence curve (JC) of a Fisher-consistent functional was introduced by F. Hampel and plays a central role in the search for robust estimators. An extension of this notion to non-Fisher-consistent functionals is proposed in order to investigate the infinitesimal robustness of more general statistics, e.g. those used in hypothesis testing. This new definition inherits many useful properties, including some on asymptotic efficiency. Functionals in two variables, arising from two-sample statistics, are treated too. Connections with Hodges-Lehmann shift estimators are discovered. One- and two-sample rank statistics illustrate the theory.